Cutaneous barrier leakage and gut inflammation drive skin disease in Omenn syndrome.

BACKGROUND: Severe early-onset erythroderma and gut inflammation, with massive tissue infiltration of oligoclonal activated T cells are the hallmark of Omenn syndrome (OS).
OBJECTIVE: The impact of altered gut homeostasis in the cutaneous manifestations of OS remains to be clarified.
METHODS: We analyzed a cohort of 15 patients with OS and the 129Sv/C57BL/6 knock-in Rag2R229Q/R229Q (Rag2R229Q) mouse model. Homing phenotypes of circulating lymphocytes were analyzed by flow cytometry. Inflammatory cytokines and chemokines were examined in the sera by ELISA and in skin biopsies by immunohistochemistry and in situ RNA hybridization. Experimental colitis was induced in mice by dextran sulfate sodium salt.
RESULTS: We show that memory/activated T cells from patients with OS and from the Rag2R229Q mouse model of OS abundantly express the skin homing receptors cutaneous lymphocyte associated antigen and CCR4 (Ccr4), associated with high levels of chemokine C-C motif ligands 17 and 22. Serum levels of LPS are also elevated. A broad Th1/Th2/Th17 inflammatory signature is detected in the periphery and in the skin. Increased Tlr4 expression in the skin of Rag2R229Q mice is associated with enhanced cutaneous inflammation on local and systemic administration of LPS. Likewise, boosting colitis in Rag2R229Q mice results in increased frequency of Ccr4+ splenic T cells and worsening of skin inflammation, as indicated by epidermal thickening, enhanced epithelial cell activation, and dermal infiltration by Th1 effector T cells.
CONCLUSIONS: These results support the existence of an interplay between gut and skin that can sustain skin inflammation in OS.

Maintenance of skin homeostasis relies on the crosstalk among epidermal keratinocytes, resident immune cells, and commensal microbiota. Breaching of the cutaneous barrier and inadequate or excessive immune response to microbial challenges potentially lead to chronic inflammation and autoimmunity. A large number of T cells normally populate the skin, providing immune surveillance. In addition to this resident population, a subset of recirculating memory T cells has been identified, contributing to widespread immune protection of distal lymphoid and cutaneous tissues., Migration of T cells into the skin requires ligands for E- and P-selectin, among which an important role is played by the P-selectin ligand CD162, also known as cutaneous lymphocyte antigen (CLA). Moreover, skin-homing T cells express the chemokine receptors CCR4 and CCR10, whose ligands—chemokine C-C motif ligands (CCL)17, CCL22, and CCL27—are upregulated in the dermal vasculature and in epidermal keratinocytes in many inflammatory conditions.5, 6, 7 Imprinting of skin-homing properties in T cells critically depends on skin-derived dendritic cells (DCs); T cells activated in skin-draining peripheral lymph nodes (PLNs) express higher levels of selectin ligands than those activated in mesenteric lymph nodes (MLNs). Conversely, the gut-homing receptors α4β7 and CCR9 are preferentially induced when T cells are activated in MLN. Thus, in general, effector T cells trafficking to the skin are devoid of gut-homing molecules and gut-tropic cells do not express skin-homing molecules. Nonetheless, evidence of plasticity in the homing phenotype has been accumulating in the context of inflammation.

Cutaneous manifestations are common in patients with primary immunodeficiency disease., Omenn syndrome (OS) is a severe immunodeficiency most often caused by hypomorphic mutations in the RAG family, leading to generation of oligoclonal T cells that expand in the periphery and infiltrate various organs, particularly barrier tissues such as skin and gut. The most distinctive clinical feature of OS is a generalized exfoliative erythroderma with onset at neonatal age, associated with alopecia, lymphadenopathy, hepatosplenomegaly, recurrent and severe infections, and failure to thrive. Skin biopsies from patients with OS reveal massive T-cell infiltration and an aberrant distribution of Langerhans cells, indicating that immune abnormalities play an important role in the skin disease pathophysiology., Patients with OS also frequently suffer from chronic diarrhea, and biopsies of the gastrointestinal tract reveal prominent T-cell and eosinophil infiltrates in the lamina propria, villous flattening in the small bowel, and cryptitis in the colon. A close association between skin disease and microbiota (not limited to skin microbiota) has been demonstrated. For instance, atopic dermatitis (AD) and rosacea are both linked to changes in the gut barrier and intestinal microbiota. Reduced microbial diversity was reported in infants with eczema., Furthermore, a link between imbalances of the intestinal microbial communities and development of skin psoriasis has been documented in mice and patients. We have recently demonstrated that intestinal microbiota play a major role in the immune dysregulation of the 129Sv/C57BL/6 knock-in Rag2R229Q/R229Q (Rag2) mouse model of OS, leading to the systemic dissemination of gut-derived proinflammatory T1/T17 cells. Whether such alterations in gut homeostasis may also influence the cutaneous manifestations of OS is not known.

Here, we provide evidence supporting a role for gut-skin axis in the pathophysiology of OS. We found that compromised skin barrier integrity in Rag2 mice results in altered microbial load and composition and dysregulated production of T cell–recruiting chemokines by activated epithelial cells. Importantly, intestinal inflammation and gut barrier leakage synergistically support the activation of skin epithelial cells and the recruitment of T cells to the skin. An increased proportion of circulating T cells expressing skin-homing receptors and elevated plasma levels of LPS and skin chemoattractants have been also documented in patients with OS, suggesting that mechanisms similar to those observed in the mouse model may also operate to induce skin inflammation in humans.

Methods

Human studies

Patient’s samples were collected with the informed consent of their parents, according to protocol 18-I-0041 approved by the National Institutes of Health institutional review board and protocols TIGET06 and TIGET09 approved by the San Raffaele Scientific Institute ethics committee. Table E1 (in the Online Repository available at www.jacionline.org) describes clinical, immunological, and molecular features of 15 unrelated patients with OS. A cohort of age-matched healthy donors (HDs), patients with acute graft-versus-host disease and patients with AD were also studied as controls. PBMCs from peripheral blood were purified on Ficoll gradient (Lympholyte cell separation media; Cedarlane Laboratories, Burlington, Ontario, Canada).

Table E1

Clinical features of RAG patients

ID patient	Gene	Mutation	Erythroderma	Colitis/diarrhea	Autoimmunity	Diagnosis
SCID-400-MX	RAG2	p. [L155P; R229W]	Yes	Yes	No	OS
SCID-417-UK	RAG1	p. [R410Q/M1006V; M435V/M1006V]	Yes	No	No	OS
SCID-418-UK	RAG1	p.[R396C; R404Q]	Yes	Yes	No	OS
SCID-419-UK	RAG1	p. [R561C; R716W]	Yes	Yes	No	OS
SCID-420-UK	RAG1	p.[M435V/R897∗; M1006V]	Yes	Infectious (rotavirus)	No	OS
SCID-421-UK	RAG1	p.[K86Vfs∗33; R559S]	Yes	Infectious (enterovirus)	ITP, hepatitis	OS
SCID-426-KW	RAG1	p. [L454Q; L454Q]	Yes	No	No	OS
SCID-454-IT	RAG1	p. [E147fs∗27; E147fs∗27]	Yes	Yes	No	OS
731-UK	RAG1	p. [R829fs∗2; K86Vfs∗33]	Yes	Yes	No	OS
769-AF	RAG1	p. [K186Sfs∗15; K186Sfs∗15]	Yes	Yes	No	OS
884-PA	RAG1	p. [Q242R/R404Q; N766I]	Yes	Yes	No	OS
SCID-375-MX	RAG2	p. [G35V; G35V]	Yes	Yes	No	OS
SCID-452-KW	RAG1	p. [R394W; R394W]	Yes	Yes	No	OS
HIE-118-LB	RAG1	p. [R404W; R737H]	Yes	No	ITP, AIHA	OS

AIHA, Autoimmune hemolytic anemia; ITP, idiopathic thrombocytopenic purpura; SCID, severe combined immunodeficiency.

Mice

The Rag2R229Q mice were previously described. The colony was maintained in a specific pathogen–free facility in heterozygosity and littermates were kept in the same cage until weaning. Both male and female mice between 8 and 12 weeks old were used for experiments. OVA transgenic OT-I and OT-II mice were obtained from Charles River Laboratories (Wilmington, Mass). The animal procedures were approved by the Istituto Clinico Humanitas, according to the Italian Institutional Animal Care and Use Committee and European Union directive for animal experiments. Details about the experimental design, in vivo procedure, and treatments are described in the supplementary Methods in this article’s Online Repository at www.jacionline.org.

ELISA tests

Quantification of mouse Ccl17 and Ccl22 and human CCL17, CCL22, and C-X-C motif chemokine ligand (CXCL)-10 were determined with ELISA kits (R&D Systems, Minneapolis, Minn), according to manufacturer’s instructions. Human LPS-binding protein was determined in the sera of patients with OS and age-matched HDs using human LBP ELISA kit (MyBioSource, San Diego, Calif), used according to manufacturing’s instructions.

Quantification of adherent bacteria to cutaneous epithelium

One centimeter of skin biopsies was collected from age- and sex-matched mice using sterile conditions and digested in lysis buffer (10 mmol/L TRIS, 100 mmol/L NaCl, 10 mmol/L EDTA, 0.5% SDS, and 0.4 mg/mL proteinase K). Total DNA was extracted with the isopropanol-ethanol method. The microbial load was determined with real-time quantitative PCR (SYBR green) using 16S and 18S ribosomal RNA gene-specific probes.

Mouse cutaneous cell preparation

Skin tissue was digested with Liberase TL (0.45 mg/mL; Roche, Basel, Switzerland) for 1 hour at 37°C. Cell suspension was washed, passed through a 100-μm cell strainer, and analyzed by flow cytometry,

Lymph flow assessment using EB

Three μL of 1% Evans blue dye (EB) solution in PBS was injected into the back of anesthetized mice using a Hamilton syringe (Hamilton Company, Reno, Nev). After 16 hours mice were sacrificed. EB was extracted from the skin biopsies of the back by incubating them at 55°C for 5.5 hours in formamide (Honeywell Fluka, Thermo Fisher Scientific, Waltham, Mass). The absorbance was measured with microplate reader at 620 nm. The dye concentration was calculated from a standard curve of EB in formamide (Sigma-Aldrich, St Louis, Mo) and is represented as absolute amount of dye remaining in the skin tissue.

Flow cytometry

Cell suspensions obtained from mouse organs were preincubated with FcγR-specific blocking antibodies (eBioscience, San Diego, Calif) and stained with the following antibodies from eBioscience: TCR-γδ (UC7-13D5), CD45 (30-F11), CD4 (RM4-5), CD8a (53-6.7), CCR9 (CW-1.2), CD103 (2E7), CD11C (N418), MHCII (M5/114.15.2), CCR4 (2G12), IL-17A (17B7) and IFN-γ (XMG1.2). Dead cells exclusion was performed with Aqua Cell Stain kit (Invitrogen, Thermo Fisher Scientific). For intracellular staining, cells were stimulated for 4 hours with phorbol 12-myristate 13-acetate (5 ng/ml; Sigma-Aldrich) and ionomycin (1 μg/ml; Sigma-Aldrich). Golgi stop (1000×; BD, Franklin Lakes, NJ) was added during the last 3 hours of stimulation. Cell fixation and permeabilization was performed using the Fixation and Permeabilization Buffer kit (eBioscience). Human cells were stained with the following antibodies from BioLegend (San Diego, Calif): CD4 (OKT4), CD8 (RPA-T8), CD45RA (HI100), CD45RO (UCHL1), CLA (HECA-452), CCR4 (L291H4). Fluorescence-activated cell sorting (FACS) data were acquired with FACSCanto II (BD) and analyzed with FlowJo software (version 9.9.6; Tree Star, BD).

Real-time PCR

Cutaneous tissues (1 cm2) were homogenized in PureZOL reagent (Bio-Rad Laboratories, Hercules, Calif) using TissueLyzer II (QIAGEN, Hilden, Germany). RNA was extracted using RNeasy Lipid Tissue kit (QIAGEN). A total of 1 μg of RNA was used to synthetize cDNA using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, Calif). Gene expression was investigated by SYBR green RT-qPCR using the CFX384 system (Bio-Rad). The housekeeping gene β-actin (Actb) ribosomal RNA was used as control. List of primers are available in Table E2 in this article’s Online Repository at www.jacionline.org.

Table E2

List of primers used in this study

Primer ID	5'-3' Sequence
Actb For	CTAAGGCCAACCGTGAAAAG
Actb Rev	ACCAGAGGCATACAGGGACA
IL-17A For	TCCAGAAGGCCCTCAGACTA
IL-17A Rev	TGAGCTTCCCAGATCACAGA
IFN-γ For	TCAAGTGGCATAGATGTGGAAGAA
IFN-γ Rev	TGGCTCTGCAGGATTTTCATG
IL-5 For	ATCCAGGAACTGCCTCGTC
IL-5 Rev	ATCCAGGAACTGCCTCGTC
IL-13 For	CCTCTGACCCTTAAGGAGCTTAT
IL-13 Rev	CGTTGCACAGGGGAGTCT
IL-33 For	TCCTTGCTTGGCAGT
IL-33 Rev	TGCTCAATGTGTCAA
TNF-α For	GACGTGGAACTGGCAGAAGAG
TNF-α Rev	TTGGTGGTTTGTGAGTGTGAG
IL-1β For	GCCCATCCTCTGTGACTCAT
IL-1β Rev	AGGCCACAGGTATTTTGTCG
CCL20 For	AACTGGGTGAAAAGGGCTGT
CCL20 Rev	GTCCAATTCCATCCCAAAAA
CXCL10 For	GCTGCCGTCATTTTCTGC
CXCL10 Rev	TCTCACTGGCCCGTCATC
CXCL9 For	CTTTTCCTCTTGGGCATCAT
CXCL9 Rev	GCATCGTGCATTCCTTATCA
16s For	GTGSTGCAYGGYTGTCGTCA
16s Rev	ACGTCRTCCMCACCTTCCTC
18s For	CTCAACACGGGAAACCTCCTCAC
18s Rev	CGCTCCACCAACTAAGAAGG
IL-23p19 For	AGCCAGTTCTGCTTGCAAAGG
IL-23p19 Rev	GGAGGTTGTGAAGTTGCTCCATG
Cramp For	AAGGAGACTGTATGTGGCAAGGCA
Cramp Rev	TTTCTTGAACCGAAAGGGCTGTGC
mBD-3 For	GTCAGATTGGCAGTTGTGGA
mBD-3 Rev	GCTAGGGAGCACTTGTTTGC
VEGF-A For	CAGGGCTTCATCGTTACAG
VEGF-A Rev	CATCTTCAAGCCGTCCTGT
TLR-2 For	AGCATCCTCTGAGATTTGA
TLR-2 Rev	GGGGCTTCACTTCTCTGCTT
TLR-3 For	TCCCCCAAAGGAGTACATT
TLR-3 Rev	GATACAGGGATTGCACCCA
TLR-4 For	GGACTCTGATCATGGCACTG
TLR-4 Rev	CTGATCCATGCATTGGTAGGT
TLR-5 For	CTGGAGCCGAGTGAGGTC
TLR-5 Rev	CGGCAAGCATTGTTCTCC
TLR-9 For	GAGAATCCTCCATCTCCCAAC
TLR-9 Rev	CCAGAGTCTCAGCCAGCAC
TSLP For	CAGCTTCGTCTCCTGA
TSLP Rev	AAATGTTTTGTCGGGGAGTG
CK5 For	CATTCTCAGCCGTGGTACG
CK5 Rev	CAGAGCTGAGGAACATGCAG
CK6 For	GTCCAGGACCTTGTTCTGCT
CK6 Rev	ATCCAGCGGGTCAGGACT
CCL22 For	CTGATGCAGGTCCTATGGT
CCL22 Rev	GGAGTAGCTTCTTCACCCAG
CCL17 Rev	TGCTTCTGGGGACTTTTCTG
CCL17 For	GAATGGCCCCTTTGAAGTAA
CXCR3 Rev	GGCATCTAGCACTTGACGTTC
CXCR3 For	GCAGCACGAGACCTGACC
Loricrin Rev	CATGAGAAAGTTAAGCCCATG
Loricrin For	GGTTGCAACGGAGACAACA
Involucrin Rev	TTGAGAGGTCCCTGAACCAC
Involucrin For	TCCTGTGAGTTTGTTTGGTC

For, Forward; Rev, reverse.

Histology and immunohistochemistry

Mouse tissue samples were formalin-fixed and paraffin-embedded. Sections (1.5 μm) were used for routine hematoxylin and eosin (H&E) analysis and staining with primary antibodies. Arbitrary gut and skin histological score was calculated in a double-blind study by a pathologist. Scoring from 0 to 3 (where 0 is normal condition and 3 is severe structural changes) was adopted for evaluating different manifestations in each tissue. Detailed description of histological and morphometric analyses are described in the supplementary Methods in the Online Repository.

In situ RNA hybridization

In situ hybridization for IFN-γ, CXCL9, and CXCL10 on formalin-fixed paraffin-embedded samples was performed using the RNAscope 2.5 HD Assay-BROWN (Advanced Cell Diagnostics, Newark, Calif) according to the manufacturer's instructions. POLR2A mRNA labelling was used as positive control for the RNA quality in samples. Staining procedure is described in the supplementary Methods in the Online Repository.

Statistical analysis

Statistical analysis was conducted using GraphPad Prism software (GraphPad Software Inc, La Jolla, Calif). Results were analyzed using the nonparametric Mann-Whitney test or the ANOVA with multiple comparisons test. Data are generally shown as mean ± SEM unless otherwise stated. Value of P < .05 was considered significant.

Results

Enhanced skin tropism of PB memory T cells in patients with OS

Erythroderma, with skin infiltration by autoreactive and oligoclonal autologous T cells, represents the clinical hallmark of OS. CLA controls the influx of memory T cells to cutaneous sites. We studied 15 patients with OS due to RAG mutations (Table E1) and 37 healthy infants (HDs) and examined CLA expression in PB T cells. In both CD4 and CD8 effector/memory CD45R0+ T cell subsets, the frequency of CLA+ cells was dramatically increased in patients with OS compared with age-matched HDs (Fig 1, A). Circulating CLA+ T cells coexpress the chemokine receptor CCR4 and rely on T-cell chemoattractant CCL22 and CCL17 to reach the inflamed skin., Indeed, CCR4 was expressed on almost the totality of CD4+ CD45RA− cells in patients with OS (Fig 1, A), and higher levels of the CCR4 ligands CCL17 and CCL22 were detected in the plasma, as compared to HDs (Fig 1, B). These features are consistent with the T2 phenotype of OS, as also indicated by elevated serum levels of IL-5 and IL-13 (Fig 1, C). Surprisingly, we found that patients with OS exhibited also significantly high plasma levels of IFN-γ and IFN-γ–inducible chemokines CXCL9 and CXCL10, as well as of IL-13, IL-5, and IL-17 (Fig 1, B and C). Furthermore, plasma levels of T1 and T2 cytokines were significantly higher in patients with OS than in patients with graft-versus-host disease and AD, 2 prototypic conditions characterized by T1 and T2 skewing, respectively (Fig 1, C). Altogether, these data indicate that the inflammation of OS is characterized by a systemic and broad signature. Consistent with this, IFNG and to greater extent CXCL9 and CXCL10 mRNA signals were readily detected in skin biopsies from patients with OS (Fig 1, D and Fig E1 in this article’s Online Repository at www.jacionline.org).

Fig 1

Preferential homing of T cells to the skin in patients with OS. A, Representative FACS plots and frequencies of CLA+ CD45RO+ T cells within the effector CD4+ and CD8+ T-cell populations isolated from PB of patients with OS (n = 13) and aged-matched HDs (n = 31). Frequencies of CCR4+ within the CD4+ CD45RA− T-cell population (n = 11-15). Plasma levels of chemokines (B) and cytokines (C) in HDs (n = 37), in patients with OS (n = 15) as well as in patients with acute graft-versus-host disease (GvHD) (n = 15) and patients with AD (n = 14), as controls of T1- and T2-mediated inflammation, respectively. D, Skin biopsies from patients with OS were investigated for IFNG, CXCL9, and CXCL10 mRNA expression using RNAscope. POLR2A mRNA labeling was used as control for the quality of the RNA on paraffin section. Values are mean ± SEM. ∗P < .05, ∗∗P < .01, ∗∗∗P < .005, ∗∗∗∗P < .001. Statistical significance determined by Mann-Whitney U test or by ANOVA. See also Table E1.

Fig E1

Elevated CXCL9 expression in skin biopsies from patients with OS. Skin biopsies from 2 patients with OS were analyzed for CXCL9 mRNA expression using RNAscope. POLR2A mRNA labeling was used as control for the quality of the RNA on paraffin section.

Skin inflammation in Rag2R229Q mice is characterized by T-cell infiltration and a T1-biased cytokine response

To investigate specifically the pathophysiology of skin disease in OS, we used the Rag2 mouse model., Typically, mutant mice develop epidermal hyperplasia, hyperkeratosis, and dermal immune cell infiltration (Fig 2, A and B). Such alterations, associated with massive signs of subcutaneous hemorrhage and PLN enlargement, were much more pronounced in mice with overt skin erythroderma (Fig E2 in this article’s Online Repository at www.jacionline.org). Immunophenotypic analysis of the skin infiltrate confirmed an expansion of mononuclear CD45+ cells, particularly of CD4+ and CD8+ T cells (Fig 2, C and D). Moreover, the cutaneous expression of the chemokine receptor Cxcr3 was increased in mutant mice (Fig 2, E). Consistent with this, CD3+ T cells isolated from the skin of Rag2 mice displayed much higher frequencies of Ifn-γ–producing cells on phorbol 12-myristate 13-acetate and ionomycin pulsing than those from wild-type (WT) control mice (Fig 2, F). Of note, the proportion of Il-17–producing cells in mutant mice was comparable to what observed in controls, despite lack of Tcrγδ+T cells, the main producers of this cytokine in the skin (Fig 2, F and G).

Fig 2

Rag2 mice exhibit severe T-cell–mediated inflammation in the skin. A, Representative skin sections from Rag2 and Rag2 mice (8-12 weeks old) stained with H&E and CD3 immunostaining. Histogram shows the inflammation score in the skin (n = 12-13). Bars = 200 μm (H&E) and 100 μm (CD3). B, Total cell count from skin suspension for g tissue (n = 7-8 mice/group). C, Frequencies of CD45+ cells in the skin suspension (n = 9-10 mice/group). D, Representative FACS plots and cumulative frequencies of CD4+ and CD8+ T cells inside the CD45+ population of skin suspension (n = 6-12 mice/group). E, Gene expression analysis of Cxcr3 in the skin tissue of Rag2 and Rag2 mice (n = 6-9 mice/group). Target mRNA was normalized to Actb mRNA. RNA contents are shown as arbitrary units (AUs). F, Representative FACS plots and frequencies of T1, T17, and T2 cells inside the CD3+ T-cell population of skin suspension (n = 6-9 mice/group). G, Representative FACS plots showing TCRγδ T cells inside the CD45+ population in the skin of Rag2 and Rag2 mice. Values are mean ± SEM. ∗P < .05; ∗∗P < .01; ∗∗∗P < .005; ∗∗∗∗P < .001. Statistical analysis was performed using Mann-Whitney U test. See also Fig E1.

Fig E2

A small proportion of Rag2 mice exhibits erythroderma-like phenotype. A, Cutaneous phenotype of Rag2 mice affected by erythroderma. B, Macroscopic appearance of dorsal skin and lymph nodes of Rag2 and Rag2 with erythroderma-like phenotype. C, Representative skin sections from Rag2 with evident skin erythroderma-like phenotype stained with H&E, KI67 (upper right; bar = 500 μm), CD3 (bottom left), and myeloperoxidase (bottom right; bar = 100 μm) immunostaining. Higher magnification shows eosinophils. D, Histogram shows the inflammation score of the skin (n = 5-6 mice/group). Values are mean ± SEM. ∗∗P < .01. Statistical analysis was performed using Mann-Whitney U test.

Rag2R229Q mice exhibit dysfunction of the epidermal barrier and increased microbial challenge

At steady state, the epidermis of Rag2R229Q mice appears thickened compared with the epidermis of WT mice (Fig 3, A). Expression of the epidermal differentiation markers, cytokeratin 5 (Ck5) and loricrin, was markedly increased (Fig 3, B). Similarly, the mRNA level of the hyperproliferative/wound-associated cytokeratin 6 (Ck6) was also augmented (Fig 3, B), overall indicating an altered keratinocyte differentiation program. Antimicrobial peptides are significantly upregulated in mouse models with altered skin barrier., We found that cutaneous expression of cathelin-related antimicrobial peptides (Cramp) and β-defensin 3 (mβ-d3) was increased in Rag2R229Q mice (Fig 3, C). Furthermore, mutant mice exhibited augmented tissue expressions of several Toll-like receptors (Tlrs) as well as of the downstream adaptor molecules, myeloid differentiation primary response 88 (MyD88) and TIR-domain–containing adapter-inducing IFN-β (Trif) (Fig 3, D and E), suggestive of disturbance in the cutaneous barrier integrity and increased microbial challenge. In line with this, skin bacterial load was increased in Rag2R229Q mice (Fig E3, A in this article’s Online Repository at www.jacionline.org), indicating a possible expansion of the overall number of bacteria compared with in their WT littermates. Notably, 16S ribosomal RNA gene sequencing revealed that bacterial taxonomic richness was generally reduced in the skin of Rag-mutant mice (Fig E3, B), with lower abundance of Propionibacterium (Padj < .028), Bacteroides (Padj < .009), and Staphylococcus (Padj < .008) (Fig E3, C) genera, which are cutaneous commensal bacteria contributing to maintenance of tissue homeostasis.

Fig 3

Altered barrier integrity and microbial load characterized the skin of Rag2 mice. A, Representative skin sections from Rag2 and Rag2 mice stained with H&E. Bar = 100 μm. B, Cutaneous gene expression analysis of keratins (Ck5-Ck6) and Loricrin in Rag2+/+ and Rag2R229Q mice (n = 8-11 mice/group from 3 experiments). Target mRNA was normalized to Actb mRNA. RNA contents are shown as AUs. C, Gene expression analysis of Mβ-d3 and Cramp in the skin tissue (n = 4-8 mice/group). Gene expression analysis of Tlrs (D) and Myd88 and Trif (E) in the skin tissue of Rag2 and Rag2 mice (n = 6-9 mice/group). Target mRNA was normalized to Actb mRNA. RNA contents are shown as AUs. F,Rag2 and Rag2 mice received a single intradermal injection of 100 μg LPS on the left and PBS on the right portion of the back. Mice were then sacrificed 24 hours later. Representative skin sections from dermal LPS-treated and untreated Rag2 and Rag2 mice stained with H&E. Bar = 200 μm. G, Fold increase of Il-1β, Ifng, and macrophage inflammatory protein 2 (Mip-2, also known asCxcl2) on skin tissue. Data are representative results of 3 independent experiments with at least 6 mice/group. Values are mean ± SEM. ∗P < .05; ∗∗P < .01; ∗∗∗P < .005. Statistical analysis was performed using Mann-Whitney U test.

Fig E3

The composition of skin microbiota differs between Rag2 and Rag2 mice. Skin tissue microbiota from adults Rag2+/+ (n = 10) and Rag2R229Q (n = 10) littermates was analyzed through 16s rRNA gene profiling. A, Quantitative PCR analysis of cutaneous adherent bacteria in Rag2+/+ and Rag2R229Q mice expressed as prokaryotic/eukaryotic DNA (ie, 16S/18S rRNA genes) ratios (n = 12 mice/group from 3 experiments). B, Scatter dot plot depicts the intrasample (α) diversity (Chao1 index) describing taxonomic richness of skin microbiota in Rag2+/+ and Rag2R229Q mice. C, A linear discriminant analysis effect size identifies the significant different abundance of bacterial taxa between Rag2R229Q mice and controls. The taxa with significantly different abundances among the genotypes are represented by colored shadows in the cladogram of the top panel and listed in the table below. From the center of the cladogram outward, kingdom, phylum, class, order, family, and genus levels are shown. Heat maps represent the relative abundance of significantly different taxa per sample (black-red heat map). Statistical analysis of data (A, C) was performed using Mann-Whitney U test. ∗P < .05.

Following the observation that cutaneous Toll-like receptor 4 (Tlr4) expression was markedly upregulated in Rag2R229Q mice, we examined further the effect of local Tlr4 activation. Mice received a single intradermal injection of 100-μg LPS on the right portion of the back and of PBS on the left and were sacrificed 24 hours later. Hemorrhage and dermal tissue necrosis were evident at the LPS injection site in almost all treated mutant mice (Fig 3, F). In particular, histopathology of skin sections revealed widespread epidermal thickening with hyperkeratosis and increased abscess formation and hemorrhage. By contrast, WT mice showed mild signs of hemorrhage and no microabscesses (Fig 3, F). We also analyzed the expression of the CK6-inducing cytokine Il-1β as well as of Ifng, involved in the pathogenesis of inflammatory skin diseases. Expression of both cytokines was strongly induced in the skin of Rag2R229Q mice in response to LPS (Fig 3, G). Moreover, the augmented expression of the keratinocyte-derived Cxcl2 chemokine correlated with a robust recruitment of innate cells to the skin of Rag2R229Q mice (Fig 3, G). Thus, Rag2R229Q mice are more susceptible to LPS-induced cutaneous inflammation.

Chronic skin inflammation sustains immune cell infiltration in Rag2R229Q mice

Chronic microbial stimulation may induce keratinocyte activation.29, 30, 31 Skin epithelial cells of Rag2R229Q mice expressed high levels of MHC class II molecules (Fig 4, A). Moreover, the expression of several cytokines, including Ifn-γ, Tnf-α, Il-23, Il-1β, Il-5, Il-13, Il-33, and thymic stromal lymphopoietin (Tslp), was significantly upregulated in the mutant skin (Fig 4, B), suggesting that these cytokines may contribute to skin inflammation, without an obvious skewing of T responses. Consistent with this, levels of both T1 (Cxcl9, Cxcl10) and T2 (Ccl20, Ccl22, and Ccl17)-related chemokine transcripts were markedly increased in the skin of Rag2R229Q mice (Fig 4, C). Levels of Ccl17 and Ccl22 were also elevated in the serum of Rag mice (Fig 4, D), correlating with a higher frequency of Ccr4-expressing skin homing CD3+ T cells detected in the PLNs, spleen, and MLNs (Fig 4, E). Overall, these data mirror what is observed in patients with OS (Fig 1) and suggest a broad inflammatory signature in this disease.

Fig 4

Characterization of cutaneous inflammatory environment in Rag2 mice. A, Mean fluorescence intensity (MFI) for MHCII expression among CD45− population of skin suspension (n = 6-7 mice/group). Transcriptional analysis of several cytokines (B) and chemokines (C) in the skin tissue (n = 6-9 mice/group from 3 experiments). Target mRNA was normalized to Actb mRNA. RNA contents are shown as AUs. D, Levels of the chemokines Ccl17 and Ccl22 in the serum of Rag2 and Rag2 mice (n = 12-13 mice/group from 3 experiments). E, Representative FACS plots and cumulative frequencies of CCR4+ inside the CD3+ T-cell population of spleen, MLN, and DLN (n = 14 mice/group from 3 experiments). Values are mean ± SEM. ∗P < .05; ∗∗P < .01; ∗∗∗P < .005; ∗∗∗∗P < .001. Statistical analysis was performed using Mann-Whitney U test. See also Fig E3. SSC-A, Side scatter.

To test whether this inflammatory environment has a role in promoting antigen-driven cutaneous T-cell migration, we adoptively transferred CD4+ or CD8+ lymphocytes from OT-II or OT-I mice into Rag2R229Q and Rag2+/+ recipients. Then, we topically immunized the ear pinna once with ovalbumin (OVA) plus cholera toxin, and after 7 days, treated mice were sacrificed. Epicutaneous challenge caused marked epidermal changes in Rag2R229Q mice, with enhanced vascularization and dermal infiltration by immune cells (Fig E4, A-C in this article’s Online Repository at www.jacionline.org). Accumulation of transgenic T cells was significantly increased within the cervical LNs (Fig E4, D and E) and skin (Fig E4, F and G) of immunized Rag2R229Q mice compared with that of controls. Overall, these findings indicate that keratinocyte activation and cutaneous inflammatory environment in mutant mice contribute to T-cell skin homing and infiltration.

Fig E4

Cutaneous inflammatory environment in mutant mice promote T-cell infiltration. A, Representative skin section from controls and CT+OVA-treated Rag2 and Rag2 mice stained with H&E and CD31 immunostaining (asterisks mark the CD31-positive vessels). Bar = 200 μm. B, Total skin counts from skin suspension (n = 5-12). C, Representative FACS plot and frequencies of CD45+ cell population of skin suspension (n = 5-12). D and E, CD4 or CD8 splenocytes from OT-II or OT-I mice were adoptively transferred into Rag2 and Rag2 recipients. Topical immunization was applied on ears once with OVA plus CT or only CT. Recipient mice were sacrificed after 7 days. Representative dot plots of donor-derived OT-II or OT-I cells within total live population of DLN suspension. F and G, Representative dot plots and frequencies of donor-derived OT-II or OT-I cells within total live population of skin suspension (n = 5-7 mice/group). Values are mean ± SEM. ∗∗P < .01; ∗∗∗P < .001. Statistical analysis was performed using Mann-Whitney U test.

Enhanced peripheral trafficking of skin-derived DCs in Rag2R229Q mice

Cell trafficking to inflamed tissues is influenced by vascular remodeling. We found that levels of the angiogenic vascular endothelial growth factor A (Vegfa) were elevated, and CD31+ endothelial cells were present at higher number and density in the skin of Rag2 mice (Fig 5, A and B). To investigate possible abnormalities of lymphatic vessels, we quantified the uptake and drainage of the lymphatic tracking dye EB after intradermal injection into the back skin. Significantly less EB remained localized within the skin of Rag2 compared with in WT mice, indicating enhanced lymphatic clearance (Fig 5, C). Hence, trafficking of antigens and immune cells from the skin of Rag2 mice could be facilitated by the expanded lymphatic network. To test this hypothesis, allophycocyanin-labeled OVA antigen was intradermally injected in the ear pinna of mice, and its uptake and transport by DCs was evaluated in the skin and draining lymph nodes (DLNs) after 6 hours. Comparable frequencies of OVA-loaded DCs were detected locally within the skin tissue of Rag2 and Rag2+/+ mice (Fig 5, D). However, consistent with the enhanced EB dye transport, a higher proportion of OVA-positive DCs was recovered in the DLNs of Rag2 mice compared with in controls (Fig 5, D).

Fig 5

Increased DC trafficking from the skin of Rag2 mice. A, Gene expression analysis of Vegfa in the skin tissue of Rag2+/+ and Rag2R229Q mice (n = 6-9 mice/group from 3 experiments). B, Representative skin sections stained with CD31. Bar = 100 μm. The evaluation of several parameters was performed by computer-assisted morphometric analysis of CD31-stained skin sections (n = 5 mice/group). C, Quantification of extracted EB from the back skin of Rag2+/+ and Rag2R229Q mice, normalized to tissue weight (g). Data are representative results of 3 independent experiments with at least 4 mice per group. D, Allophycocyanin-labeled OVA antigen was intradermally injected in the ears of Rag2+/+ and Rag2R229Q mice, and its uptake by DCs was evaluated in the skin and DLNs after 6 hours. Representative dot plots and frequencies of OVA-loaded DCs from skin (CD11c+ MHCII+) and DLNs (CD11c+) of OVA-injected Rag2+/+ and Rag2R229Q and negative control (not injected). Data are representative results of 3 independent experiments with at least 5 mice per group. Values are mean ± SEM. ∗P < .05; ∗∗P < .01; ∗∗∗P < .005. Statistical analysis was performed using unpaired Student’s t test and Mann-Whitney U test. CTL, Control.

Skin disease in Rag2R229Q mice is exacerbated by intestinal inflammation

The aforementioned results indicate that endogenous Tlr4 ligands could have a role in cutaneous inflammation of Rag2 mice. We previously reported high levels of endotoxin (LPS) in the circulation of Rag2 mice correlating with gut dysbiosis and intestinal inflammation. Increased levels of LPS-binding protein, marker of microbial load and translocation, were detected also in sera of patients with OS (Fig 6, A), with the highest levels found in patients exhibiting intestinal inflammation and profuse diarrhea. To evaluate whether high serum endotoxin may contribute to skin inflammation in Rag2 mice, we analyzed the cutaneous responses to systemic LPS. Mice received a single injection of 100 μg LPS intraperitoneally and were sacrificed 24 hours later. Histological examination of skin biopsies showed worsening of inflammation over baseline condition in Rag2 mice, with increased epidermal thickening, hyperkeratosis, occasional abscesses, and more prominent dermal infiltrates (Fig 6, B and C). Signs of inflammation were evident also in the skin of LPS-treated Rag2 mice, similar to those observed in untreated Rag2 mice (Fig 6, C). Furthermore, on LPS challenge, levels of Tlr4 and Ifng transcripts were significantly upregulated in Rag2 mice and, to a lesser extent, in Rag2 mice (Fig 6, D), indicating that systemic endotoxin can play a role in the cutaneous inflammation of Rag2 mice. Skin homeostasis was not markedly altered in mutant mice at weaning, likely reflecting the reduced LBP levels compared with LBP levels in adult mice and the time of exposure to endotoxin (Fig E5 in this article’s Online Repository at www.jacionline.org).

Fig 6

Systemic LPS triggers cutaneous inflammation. A, LPS-binding protein (LBP) concentration in the sera of patients with OS and HDs (n = 8-9 from 2 experiments). B,Rag2 and Rag2 mice received a single injection of 100 μg LPS intraperitoneally and were sacrificed 24 hours later. Representative skin sections from systemic LPS-treated and untreated Rag2 and Rag2 mice stained with H&E. Bar = 200 μm. C, Histogram shows the inflammation score in the skin of n = 6-11 mice/group from 2 experiments. D, Fold increase of Ifng and Tlr4 on skin tissue (n = 6-8 mice/group from 2 experiments). E, Representative small intestinal and skin sections from Rag2 mice with erythroderma stained with H&E and CD3 immunostaining. Bar = 200 μm. F, Correlation between intestinal and skin inflammation scores of Rag2 mice manifesting or not evident cutaneous manifestations (black dots). The Spearman r value is indicated in the graph (n = 10 mice). Values are mean ± SEM. ∗∗P < .01. Statistical analysis was performed using Mann-Whitney U test.

Fig E5

Analysis of skin homeostasis in young Rag2R229Q mice. A, Representative skin section from 20-days-old Rag2 and Rag2 mice stained with H&E. Bar = 200 μm. B, Cumulative frequencies of CD45+, CD4+, and CD8+ T cells infiltrating the skin. C, Gene expression analysis of Ifng and Cxcl10 in the skin tissue of Rag2 and Rag2 mice (n = 6-9 mice/group). Target mRNA was normalized to Actb mRNA. RNA contents are shown as AUs. Plasma levels of (D) CCL17 and CCL22 and (E) LPS-binding protein (LBP). Values are mean ± SEM. ∗P < .05; ∗∗P < .01. Statistical analysis was performed using Mann-Whitney U test.

Moreover, we observed that Rag2 mice in which the skin inflammation was more pronounced were those in which intestinal disease was also more severe (Fig 6, E and F). To determine the potential role of the gut-skin axis in the disease pathophysiology, we boosted intestinal inflammation in Rag2 mice using chronic dextran sulfate sodium salt (DSS) treatment. Mice were sacrificed after 4 weeks. DSS-treated Rag2 mice manifested worsened skin inflammation, as indicated by enhanced epidermal thickening, hyperkeratosis, and increased number of CD3+ T cells infiltrating the dermis (Fig 7, A). Furthermore, immunostaining for CD31 demonstrated enhanced cutaneous vascularization in DSS-treated Rag2 mice (Fig 7, B). By contrast, DSS treatment did not induce similar skin abnormalities in Rag2 mice (Fig E6, A and B in this article’s Online Repository at www.jacionline.org). FACS analysis of cutaneous cell suspensions confirmed that the proportion of skin-infiltrating CD4+ T cells and CD8+ T cells was increased by 2- to 4-fold in DSS-treated Rag2 mice (Fig 7, C). Moreover, the frequency of circulating CD4+ Ccr4+ T cells was significantly increased in DSS-Rag2 mice (Fig 7, F), with a significant proportion coexpressing also the gut-homing receptor Ccr9 (Fig 7, F), thus suggesting that memory T cells in Rag2 mice may recirculate between skin and intestinal mucosa. Cutaneous levels of Cxcr3 and Ifng transcripts were upregulated in DSS-Rag2 mice (Fig 7, D), leading to enhanced epithelial expression of MHC-II molecules and keratinocyte hyperactivation (Fig 7, E).

Fig 7

Boosted intestinal inflammation worsens skin inflammation in Rag2 mice. A, Colitis in Rag2R229Q mice was induced by cyclic DSS administration. Representative skin sections stained with H&E and CD3 immunostaining. Bars = 200 μm (H&E) and 100 μm (CD3). Higher magnification shows eosinophils. Bar = 50 μm. Histogram shows the inflammation score in the skin (n = 12-15 mice/group from 3 experiments). B, Representative skin sections from untreated and DSS-treated Rag2mice stained with CD31 immunostaining (asterisks mark the CD31-positive vessels). Bar = 100 μm. Correlation between cutaneous inflammation and average vessels size of untreated Rag2R229Q (black dots) and DSS-treated Rag2R229Q (blue dots). The Spearman r value is indicated in the graph (n = 5 mice/group). C, Fold change of the CD4+ and CD8+ T-cell frequencies (n = 6-9 mice/group). D, Fold change of cutaneous Cxcr3, Cxcl10, Ifng, Ccl20, Tslp, and Tlr4 expressions (n = 6-9 mice/group). E, MFI for MHCII expression among CD45− population of Rag2 and DSS-treated Rag2 skin suspension (n = 5-6 mice/group). F, Representative FACS plots and cumulative frequencies of splenic Ccr4+, Ccr9+, and Ccr4+Ccr9+ CD4+ T cells (n = 6-11 mice/group from 3 experiments). Dotted line indicates WT averages. Values are mean ± SEM. ∗P < .05; ∗∗P < .01; ∗∗∗P < .005. Statistical significance determined by Mann-Whitney U test. See also Fig E4.

Fig E6

Cutaneous response to DSS treatment in Rag2 mice. A, Representative skin sections from untreated and DSS-treated Rag2 mice stained with CD31 immunostaining (asterisks mark the CD31-positive vessels). Bars = 100 μm. B, Correlation between cutaneous inflammation and average vessels size of untreated Rag2+/+ (black dots) and DSS-treated Rag2+/+ (blue dots). The Spearman r value is indicated in the graph (n = 5 mice/group).

Finally, we investigated whether the mechanism for endotoxin-driven skin disease exacerbation is applicable to mice not displaying preexisting inflammatory bowel disease. To this end, we tested the effect of acute DSS in a mouse model of AD. Experimental AD was induced by daily topical application of the vitamin D3 analog MC903 on the ears. At day 8, groups of mice received for 7 days water with or without 2% DSS, followed by normal drinking water for an additional 2 days, and then euthanized. Plasma LBP was significantly augmented in mice receiving DSS administration (Fig E7, A in this article’s Online Repository at www.jacionline.org), which caused increased swelling, reddening, and scaling of the skin in the treated mice (Fig E7, B and C). Microscopic investigation revealed a stronger dermal inflammatory cell infiltration with epidermal hyperplasia and hyperkeratosis in DSS-treated AD mice (Fig E7, D). Consistent with increased inflammation, the expression of T1 and T2 inflammatory markers was augmented (Fig E7, E).

Fig E7

Effect of DSS-induced intestinal inflammation on experimental model of AD. Mice were treated with MC903 for 8 days to induce AD-like dermatitis. Starting on day 8, a set of mice received 2% DSS water for 7 days, and then returned to normal drinking water for additional 2 days. A, Plasma LBP on day 17 is shown. B, Day 17 gross phenotype of AD mice treated or not with oral DSS. Bar graph shows the difference in ear thickness between day 17 and day 8 (n = 5-10 mice/group). C, Representative ear section from AD and AD+DSS mice stained with H&E (left; bar = 500 μm) and CK5 immunostaining (right; bar = 200 μm). Histogram shows the inflammation score in the ear skin (n = 5-10). D, Gene expression analysis of the ear tissue (n = 5-10 mice/group). Target mRNA was normalized to Actb mRNA. RNA contents are shown as AUs. Values are mean ± SEM. ∗P < .05. Statistical analysis was performed using Mann-Whitney U test.

Collectively, these results suggest that a leaky gut plays an important role in priming the skin disease for a hyperinflammatory response to systemic LPS.

Discussion

The cutaneous hallmark of OS is a severe exfoliative erythroderma, resulting in epidermal barrier defect and subsequent risk of life-threatening bacterial infections. The aberrant expression of skin-homing molecules CLA and CCR4 on T cells from patients with OS, together with the high CCL17/CCL22 serum levels, indicates that they are poised to migrate to the inflamed skin and have pathogenic potential. Along with high serum levels of IL-13 and IL-5, these observations may also sustain the notion that OS is a T2-driven disease. Surprisingly, we detected high levels also of IFN-γ, CXCL9, and CXCL10 in the serum of patients with OS. Similar results have been demonstrated both in the skin and in circulation of Rag2 mice, indicating that OS is not a pure T2-driven disease, but rather manifests a broad inflammatory signature, with a prominent role also for T1-mediated responses. Many studies reported promiscuous expression of chemokines in the serum of patients with cutaneous inflammation, proposing that the immunological phenotype in T-cell–mediated skin diseases can shift toward a T2/T1/T17 mixed response as the duration of the disease increases. Furthermore, although expression of CCR4 is more commonly associated with T2 responses, it has also been observed on T17 cells, suggesting that the chemokines activating this receptor may have a broader impact on the inflammatory response than simply recruiting T2 cells. Indeed, CCR4+ memory T cells also contain non-T2 populations expressing CXCR3 and IFN-γ and CCL17 and CCL22 can be released not only by leukocytes, but also by activated dermal endothelial cells and keratinocytes, in response to both T1 and T2 cytokines. Both CCR4 and CXCR3 antagonists are available for the treatment of allergic disease and psoriasis, and may represent a novel therapeutic target also in OS.

Increased expression of Tlr4 and downstream signaling molecules Myd88 and Trif in the skin of Rag2R229Q mice correlated with hyperinflammatory cutaneous responses on topical or systemic administration of LPS. Together with our previous demonstration of inflammatory bowel disease–like disease, altered gut microbiota, and elevated LPS serum levels in Rag2R229Q mice, these findings suggest that endogenous Tlr4 ligands play a role in the maintenance of keratinocyte activation and perpetuation of local inflammatory response. Consistently, DSS treatment markedly worsened cutaneous inflammation in Rag2R229Q mice. Chronic gastrointestinal diseases, such as inflammatory bowel disease or celiac disease, are often associated with skin inflammation., Likewise, marked changes in gut barrier and in intestinal microbiota are observed in AD and patients with rosacea. However, it cannot be excluded that inflammation in other organs may also contribute to increase the severity of skin inflammation in OS.

High cutaneous expression of antimicrobial peptides and Tlrs in the Rag2R229Q mice is indicative of important alterations of skin barrier integrity and microbial challenge. Changes of the cutaneous microbiota have been correlated with several skin disorders.47, 48, 49 Here we have shown that Rag2 mice manifested a selective shift in resident skin microbiota with decreased abundance of commensal bacteria belonging to Actinobacteria and Firmicutes phyla. The loss of some protective bacteria genera (Streptococcus, Staphylococcus, and Propionibacterium) might render the skin of Rag2 mice more susceptible to development of local inflammation. These data indicate the importance of performing similar studies also in patients with OS, although the frequent use of immunosuppressive agents and antibiotics in these patients may affect interpretation of the results.

Similar to patients with OS, an increased proportion of Ccr4-expressing CD4+ T cells was detectable in lymphoid organs of Rag2 mice, indicating that these cells are poised to migrate to the skin. Interestingly, we observed coexpression of skin-homing (Ccr4) and gut-homing (Ccr9) expression by a subset of splenic T cells in DSS-treated Rag2R229Q mice, suggesting that these cells might be potentially able to recirculate between gut and skin. Plasticity in the migratory property of immune cells has been suggested. Particularly, cutaneous exposure to food antigens can reprogram LN effector T cells expressing gut-homing receptors to express skin-homing receptors, thereby eliciting skin inflammation. Tissue-derived DCs have a key role in presenting local antigens and imprinting T cells with specific homing properties. Trafficking of skin-derived DCs to DLNs is increased in Rag2 mice, correlating with abundant lymphatic network. Moreover, we previously described an augmented population of migratory CD103+ DCs in the spleen of Rag2 mice, suggesting that reprogramming of T cell homing by tissue-derived DCs may also occur in secondary lymphoid organs other than DLNs. Intriguingly, skin-derived and antigen-bearing DCs were also identified in MLNs. Hence, breaches of skin and gut barriers in Rag2 mice may promote systemic translocation of microbial antigens and induce activation of T cells expressing homing receptors for skin and/or gut. In summary, we have demonstrated that the cutaneous manifestations of OS are characterized by a broad inflammatory signature and can be severely influenced by gut inflammation, reinforcing the notion of an interplay between skin and gut in the pathophysiology of the disease.

A broad T1/T2/T17 inflammatory signature is observed in the blood and skin of patients with OS and in the murine model.

Systemic LPS injection or boosting colitis in OS murine model aggravates cutaneous T-cell infiltration and skin inflammation.

High level of serum endotoxin is detected in patients with OS.